在本篇研究當中,首先我們分析了離散時間小波轉換係數的振幅上限值,然後應用比分析結果到一個以零樹為主的編碼器上。我們的分析方式是以數學歸納法為基礎。藉由分析的結果,一旦小波濾波器被選定,我們即可知道各個頻帶係數的振幅上限值。實驗結果顯示,運用此分析結果到一個以零樹為主的編碼器上,我們可以減少將近90%的計算量在尋找所有條數的最大振幅上。
In this study, we first analyze the upper bound of magnitudes of the discrete wavelet transform (DWT) coefficients, and then we apply the analysis result to a zerotree-based encoder. The analysis is based on the mathematical induction. By using the analysis result, the upper bound of magnitudes of the transform coefficients in each band is known once the wavelet filter is chosen. Experimental results show that by using the analysis results to a zerotree-based encoder we can reduce the amount of computation of finding the largest magnitude of all of the transform coefficients to near 90% of that of a full search.